View Single Post
Old 13th Oct 2017, 11:30 am   #121
Craig Sawyers
Dekatron
 
Craig Sawyers's Avatar
 
Join Date: Nov 2010
Location: Oxford, UK.
Posts: 4,941
Default Re: Audiophoolery. 'Cable Break In' - I never knew that!

Quote:
Originally Posted by Radio Wrangler View Post
We don't often disagree, Craig, but I think you've missed something.

The full Zobel network on the output of an amplifier consists of two sections. One is a series R-C pair which shunt the output of the amplifier. Together these provide a nice, resistive load to the amplifier at frequencies above audio, without burdening it significantly in the audio range. This is nice but isn't terribly important with well designed amplifiers.

The second part of the full Zobel network is a parallel indictor-resistor in series with the output (after the Zobel shunt). The job of this part is to prevent capacitive loads (exactly like those woven speaker cables) from creating excess phase shift in the feedback loop by the load capacitance and the output impedance of the output transistors creating an additional pole in the loop. The amplifier is isolated at RF from the cable/crossover capacitance and it sees the series resistor instead. This takes the cable/crossover C out of the loop.

Some high faluting amps of the nineteen eighties eschewed the series path Zobel network (presumably on religious grounds) and were renowned for taking off at RF when confronted with high-C interwoven cables. Smoke and large bills followed.

So high-C speaker cable isn't an alternative to the full Zobel, it is something which needs the full Zobel.

The business of running feedback amplifiers and feedback controlled voltage regulators into capacitive loads is a classic problem in design. The maths of loop stability scare a lot of people, but the problem is real.
We still agree. But the amp Zobel and series inductor/resistor are rarely mounted hard to the output connections (they ought to be of course). The exposed wiring between the terminals and the output network itself is an RF problem - a few centimetres of exposed cabling is an antenna at mobile phone frequency. So OK - twist the wires tightly on the way to the output terminals....

A cylindrical inductor is also a problem, since it is an RF antenna. At least one audio amp designer uses an air cored toroid for the inductor to overcome, at least to first order, this issue.

And yes - loop stability is a big issue. Get it even mildly wrong and you can put an HF pole high Q at high hundreds of kHz to low Mhz that is lurking there ready to oscillate with the wrong cable and destroy the output devices.

Lack of Zobels etc is alive and well - it was not just a feature of the 80's. Denis Morecroft's power amps have no output parts and no protection circuits (and are mounted in plastic boxes). And you have to use his speaker cables - two widely spaced wires - because a capacitive load is anathema to his designs http://www.dnm.co.uk/intro.html

Craig

Last edited by Craig Sawyers; 13th Oct 2017 at 11:39 am.
Craig Sawyers is offline   Reply With Quote